Common Questions
Frequently Asked Questions
Everything you need to know about how the index works, where the data comes from, and what the score means.
The Index
What is the Takeover Tracker?
It’s a AI Takeover Index (0–100%) that measures how much of the economy’s work AI can perform today. It scores ~20,000 occupational tasks for AI capability, weights them by employment, and produces a single transparent number grounded in real task-level data.
What does the score actually mean?
0% represents the pre-AI baseline (circa 2019). 100% would mean human white-collar labor offers no cost advantage over AI. The current estimate sits around 20%, reflecting established AI assistance and the beginning of entry-level task automation.
How often does the score update?
The AI Takeover Index updates when new AI capability assessments are run or employment data changes. Per-occupation risk scores are refreshed periodically as new task-level data becomes available.
Can the score go down?
Yes. If AI capability scores decrease for key tasks, or if employment shifts toward less-automatable occupations, the index can decrease. EMA smoothing prevents wild swings in either direction.
Methodology
Where does the data come from?
The AI Takeover Index is built on O*NET task data (~20,000 tasks across 1,100+ occupations), Anthropic’s Economic Index (real-world AI usage observations), and BLS employment statistics. Per-occupation scores also use Gemini AI estimation for tasks without observed data. Every source is documented on the Methodology page.
How is the score calculated?
Each task is scored for AI capability (0–100), weighted by how much time workers spend on it. Occupation scores aggregate task-level risks with protective factor adjustments. The headline index weights all occupations by employment. The full formula is published in our methodology.
What stops the score from being inflated by AI hype?
The AI Takeover Index is entirely task-grounded — no news sentiment, CEO predictions, or social media buzz. It moves only when AI capability scores or employment data change. This makes it inherently hype-resistant by design.
Is the methodology open?
Completely. Every data source, weight, formula, and AI prompt is published. The scoring rubric is editable and versioned. Anyone can audit, critique, or reproduce the results.
Features
Can I look up specific occupations?
Yes. We have displacement risk scores for 59,000+ jobs (1,100 base occupations plus specializations) based on O*NET task and attribute data. You can search by job title and see task-level breakdowns of AI exposure.
Can I track industries?
Yes. Browse industries, view risk scores, and track the ones relevant to you from your account dashboard.
Is there a personal risk assessment?
Yes. On any job page, sign in and toggle to “Your Score” to personalize the risk assessment with 7 quick questions about your experience and environment. Your answers are saved globally and applied across all job pages.
Is Takeover Tracker free?
Risk scores are free for all 59,000+ jobs, including impact summaries, wage data, protective factors, and basic task previews. Premium features — full task breakdowns, AI capability evidence, career transitions, PDF reports, and personalized action plans — are available from $8.99/mo per occupation.
Technical
What AI model powers the analysis?
Google’s Gemini (configurable model, currently Gemini Flash). The primary AI capability data comes from Anthropic’s Economic Index (real-world Claude usage observations). For tasks without observed data, Gemini estimates capability. The daily signal pipeline uses self-consistency sampling (3x median) for stable results.
Is there an API?
A public API is in development. Check the API Documentation page for current endpoints and usage details.
Can I contribute or give feedback?
Absolutely. The project is open to feedback. Visit our Contact page to get in touch, or check the changelog for recent updates.
Want the full picture?
The Methodology page documents every weight, formula, data source, and prompt used to produce the score.